Open Access
Add to BibSonomyAbstract
By using a 200GHz AWG channelized ASE source in connection with a saturable semiconductor optical amplifier (SOA) based noise blocker as the injecting source at the remote node in front of the local optical network units (ONUs), we demonstrate the spectrum-sliced ASE transmitter with greatly suppressed intensity noise performance in WDM-PON network. Such channelized SOA filtering technique effectively reduces the relative intensity noise of the ASE source by at least 4.5 dB. The low-noise WRC-FPLD transmitter improves its extinction-ratio (ER) from 8.9 to 9.6 dB and signal-to-noise ratio (SNR) from 5.9 to 6.3 dB. In comparison with broad-band ASE injection-locked WRC-FPLD transmitter at same power, there is an improvement on receiving power penalty (ΔPReceiver) by 2 dB at BER 10−9 in back-to-back case, and the receiving power of BER 10−9 can achieve −24 dBm even after 25km fiber transmission. With additional AWG filtering, the intraband crosstalk effect between the upstream transmitted data and the reflected ASE signal is significantly reduced by 6.3dB. The compromised effects of ER and SNR on BER performance are also elucidated via the modified SNR model for the WRC-FPLD under ASE injection induced gain-saturation condition. The ΔPReceiver/ΔSNR of 8.89 at same ER condition is more pronounced than the ΔPReceiver/ΔER of 3.17 obtained under same SNR condition, indicating that the SNR plays a more important role than the ER on enhancing the BER performance.
©2010 Optical Society of America
1. Introduction
The high bit-rate transmission of the wavelength-division-multiplexing passive optical network (WDM-PON) is demanded due to the growing population in future broadband optical access networks. Versatile cost-effective light sources were emerged to concurrently approach the colorless injection-locking capability. A typical solution is the FPLD injected by broadband ASE light source, however, its strong mode-extinction feature under the externally broadband ASE injection not only opposes the wavelength independent criterion but also constrain the broadband gain spectrum requirement [1,2]. Recently, an AR-coated weak-resonant-cavity Fabry–Pérot laser diode (WRC-FPLD) with moderate front-facet reflectance is proposed to perform the wavelength independent operation [3]. On the other hand, the ASE source inherently suffers from large intensity noise (IN) caused by spontaneous-spontaneous beat noise [4,5], such that the spontaneous–spontaneous beating noise injects into the FPLD, which degrades the signal-to-noise ratio (SNR) and causes the penalty in receiving power for obtaining up-stream transmitted data with sufficiently low bit-error-rate (BER). There has been an approach to reduce ASE intensity noise by using a gain-saturated semiconductor optical amplifier (SOA) [4–6] to filter the spectral-sliced ASE source before injection into an FPLD [7]. On the other hand, another major reason leading to the limitation on transmission bit-rate up to 2.488 Gbit/s in such a DWDM-PON is the intra-band crosstalk, which originates from the interfered effect between the transmitted data and the reflections from the facets of array waveguide grating (AWG) and feeder fiber patch-cord [8]. The crosstalk of optical reflection caused by Rayleigh scattering, bad splicing connections, and ASE intensity noise from spontaneous-spontaneous beat noise strongly affects the signal performance and network capacity. In this work, we propose a novel DWDM-PON architecture consisting of an AWG channelized and SOA filtered ASE source at the remote node in front of the local optical network units (ONUs), which is employed to inject the WRC-FPLD transmitter with front-face reflectance of 1% for increasing the direct modulation bandwidth and enabling the OC-48 transmission. Adding the AWG and SOA significantly reduces the optical reflection and intensity noise of the injected ASE source when comparing with the conventional architecture. When directly modulating the WRC-FPLD at 2.488 Gbit/s in such a DWDM-PON with AWG bandwidth of 200 GHz, the proposed system benefits from a penalty of −1.8 dB on receiving power at BER of 10−9. Moreover, the correlation between SOA operation condition and the SNR or extinction ratio (ER) are analyzed. The effects of ER and SNR on the BER performance are also elucidated via the modified SNR model for the WRC-FPLD under ASE injection induced gain-saturation condition.
2. Experimental setup
Most of previous works established the DWDM-PON system with broadband injecting source at central office, as shown in Fig. 1(a) . In contrast, the Fig. 1(b) schematically illustrates a modified DWDM-PON system constructed by allocating the spectral-sliced and SOA-filtered ASE source at remote node for injection-locked WRC-FPLDs at all ONUs. The external injection-locking source is an Erbium-doped fiber amplifier (EDFA) based broadband source, which filtered by 200 GHz AWG and then pass through SOA to inject the WRC-FPLD. The WRC-FPLD exhibits a threshold current of about 25 mA, a longitudinal mode spacing of 0.6 nm, the back and front facet reflectivity of 100% and 1%. The maximum injection power of WRC-FPLD is limited as −3 dBm due to the injection power too large would damage the end-face AR coating of WRC-FPLD.
Fig. 1 (a). The configuration a conventional ASE injecting transmitter wavelength independent operation WDM-PON. (b)Configuration of the DWDM-PON with WRC-FPLD injection-locked by the source of ASE through SOA at the end of remote node.
In our experiment, the operating current of the WRC-FPLD is detuned between 28 and 40 mA, correspond to a change from 1.1 Ith to 1.5 Ith. Latter on, the WRC-FPLD transmitter is directly modulated at 2.488 Gbit/s with a NRZ pseudorandom bit sequence (PRBS) pattern length of 223-1. To suppress the intensity noise of ASE, the SOA has to be operated at high bias and a large input power is necessary to ensure gain saturation. The maximum input power and biased current of SOA are limited at −3 dBm and 350 mA, respectively, to prevent of the SOA from damage.
3. Results and discussions
3.1 Effects of SOA and WRC-FPLD operating conditions on the up-stream transmitted data performances (BER, SNR, and ER)
The power-current characteristics of the WRC-FPLD under different injection levels are also shown in Fig. 2 , which illustrates a reduction of threshold current by 6 mA with the external injection power increasing by 9 dB. Note that when using the SOA filtered and spectrum-sliced ASE source to injection-lock the WRC-FPLD, the threshold current is slight increased as compared to that of the WRC-FPLD injected by spectrum-sliced ASE without SOA filter. The SOA introduces additional phase modulation-induced chirp [9] and four-wave mixing (FWM) effect, which inevibly broadens the spectra output of the AWG channelized ASE. The broadened ASE spectrum remains the output power constant but attenuates the photon density at specific wavelength, such that the equivalent optical intensity induced threshold current of WRC-FPLD will be less affected accordingly. The WRC-FPLD output power can be increased by at least 3 dB with its threshold current decreasing from 26 to 17 mA under enlarged ASE injection. It is also important to improve the SNR and ER of the WRC-FPLD transmitted data under the SOA filtered and AWG-sliced ASE injection with increasing power level.
Fig. 2 Power-current curves of WRC-FPLD operated without and with injection power of −12 and −3 dBm.
By externally modulating AWG spectrum-sliced and SOA filtered ASE, the SNR of the data stream at 2.488 Gbit/s reveals an increasing trend with enlarging bias of SOA and increasing power of ASE. Without the SOA based noise blocker, the transmitted data exhibits a SNR as low as 2.3 dB. The SNR of the transmitted data can be improved by 0.8 dB if the SOA bias increases from 200 to 350 mA. A further increment up to 1.2 dB can be done by increasing ASE power up to −3 dB. When SOA is operated at 350 mA and the input power is −3 dBm, the SNR of the transmitted data greatly improves from 2.3 to 6.8 dB, as shown in Fig. 3 . Adding the SOA based noise blocker essentially improves the SNR of upstream data from 6.0 to 6.3 dB. In general, the SNR of the WRC-FPLD is correlated with the integral of its relative-intensity-noise (RIN) spectrum by [10]
Fig. 3 SNR of AWG-sliced ASE without (pink-dotted) and with SOA based filter at different biased currents.
Technically, the RIN suppression can be done by adding a SOA after the spectrum-sliced ASE channel prior to injection-lock the WRC-FPLD. The RIN suppression ratio in decibel unit is theoretically given by [6]
where Γ is the mode confinement, g(N) is the gain coefficient, αint is the internal loss, J is the injected current density, d is the active layer thickness of the SOA, A is the active region area, q is the electronic charge and τc is the carrier lifetime. ω is the angular frequency at which the perturbing signal varies, g' is the differential gain coefficient, P0, and N0 are the time-averages of the output power and equivalent carrier density, and ΔP(ω) and ΔN(ω) stand for the noise. Equation (2) shows that P0, Γg(N0) and Γg’ must be extremely large for obtaining high SNR. For a given SOA, P0 and Γg(N0) can be increased by enlarging the input power Pin and the biased current Ib. In experiment, we have measured the RIN spectra of the spectrally sliced ASE source without and with additional SOA based noise blocker, and the WRC-FPLD under free-running and injection-locking conditions, as shown in Fig. 4 .The original RIN of the spectrally sliced ASE source exhibits a RIN of −135 dB/Hz below 6 GHz. By adding the SOA based noise blocker, it is clearly seen that the RIN of the ASE source decreases by 2 dB at f<3 GHz. With such an SOA filtered ASE injection, the WRC-FPLD greatly reduces its RIN from −133 dB/Hz (free-running case) to −138 dB/Hz. These results have elucidated that the WRC-FPLS essentially reduces its RIN by at least 4 dB after injection-locking with the SOA filtered ASE source, such that the transmission performance of the WRC-FPLS can further be improved. The BER of the upstream transmitted data from WRC-FPLD injection-locked by the ASE with SOA filter at different biased currents are analyzed in Fig. 5 . When the SOA biased current is enlarged from 81 to 350 mA, the BER error floor is decreased from 10−9 to 10−11, reflecting that the RIN of the WRC-FPLD is significantly diminished under SOA filtered ASE injection.
Fig. 5 WRC-FPLD up-stream BER under the injection of AWG-sliced ASE with SOA filter at different biases.
The WRC-FPLD biased current dependent BER analysis at bit rate of 2.488 Gbit/s is shown as Fig. 6 . The BER performance is greatly improved as the WRC-FPLD biased current increases from 28 to 36 mA (i.e., from Ith to 1.4 Ith), however, which turns to be degraded when the WRC-FPLD bias further increases to 40 mA or larger. At optimized bias of 36 mA, the receiving power required for the up-stream transmitted data is −26 dBm at BER of 10−9. At lower biased condition, the SNR of WRC-FPLD transmitted data predominates the BER performance, and there is a positive contribution of biased current for enhancing the SNR as well as BER. Nonetheless, the ER becomes more pronounced than SNR to degrade the BER performance as the WRC-FPLD biased current increases beyond 36 mA, while the ER significantly degrades at higher biased current to introduce a positive power penalty of 4 dB at BER of 10−9.
Fig. 6 BER and corresponding eye-diagrams of the SOA filtered ASE injection-locked WRC-FPLD at different biases of (a) 28 mA, (b) 32 mA, (c) 36 mA, (d) 42 mA.
Figure 7 illustrates the SNR and ER as a function of the WRC-FPLD biased current. Apparently, the SNR increases 5.1 to 6.4 dB as the WRC-FPLD gain enlarges when increasing its bised current from 28 to 40 mA. However, the lowest biased current (28 mA) can provide the highest ER of 13.5 dB for the WRC-FPLD transmitted data, whereas the ER oppositely degrades to 8.2 dB at biased current of 40 mA. In experiment, we observe that the effect of the ER on BER performance can only be neglected if the ER of the WRC-FPLD tansmitted data is larger than 9 dB, and SNR becomes dominated at highly biased current. Theoretically, the Q parameter which decides the lowest BER achieved in a communication system can be described in terms of Q = [(ER-1)/(ER + 1)](M⋅SNR)0.5 with M denoting the gain of the optical receiver. With increasing WRC-FPLD bias, the factor of (ER-1)/(ER + 1) is decreased from 0.89 to 0.74, where the (SNR)1/2 is only improved from 1.6 to 2.1. That is, the Q parameter behaves like a nonlinear function of WRC-FPLD bias, as shown in Fig. 8 .
3. 2 Theoretical and experimental analyses on the ASE injection power dependent SNR, ER, and BER of WRC-FPLD up-stream data
To investigate the effects of ASE injecting power and reflected ASE signal on the WRC-FPLD transmitted upstream data, we further consider the SNR and ER of the injection-locked WRC-FPLD under gain-saturation condition. Originally, the SNR of WRC-FPLD under external injection is given by [10]
where Pinj is the injection power, G is the optical gain, Δf is the detector bandwidth, Ssp is the spectral density of spontaneous-emission induced noise. Due to the gain saturation of the WRC-FPLD under ASE injection (occurred at Psat > −7 dBm), the power gain of WRC-FPLD is rewritten asHowever, the power gain of WRC-FPLD is greatly modified under the ASE external injection-locking condition as [11–13]
where G0 is the WRC-FPLD gain at free-running case, ΔG is the gain variation due to the SOA filtered and AWG-sliced ASE injection, τp is the photon lifetime, Rsp denotes the spontaneous emission rate, Pm denotes the total power at the injection-locked mode, βc denotes the linewidth broadening factor, kc denotes the coupling coefficient of the WRC-FPLD for external injection, and Pinj is the externally ASE injected power.where I and Ith are the bias and threshold currents of WRC-FPLD, Ntr is the transparent carrier number, g’ is the differential gain coefficient for ΔN denoting as the variation of carrier numbers, ηd is the differential quantum efficiency, hν is the energy per photon, ηi denotes the internal quantum efficiency, and τc denotes the carrier lifetime. When the AWG-sliced ASE injection-locked WRC-FPLD is operated in linear-gain region, the Eqs. (3) and (5) are combined to give the SNR asIf the WRC-FPLD is operated at gain-saturated condition, the SNR is modified using Eqs. (3)-(6), as described by
As a result, the measured SNR and ER of the back-to-back transmitted WRC-FPLD up-stream data are shown in Fig. 9. The SNR linearly increase with injection power, but saturate at high injection level. At the same time, the P-I curve of WRC-FPLD illustrates a decreasing trend on its threshold current with increasing ASE injection power. Note that the power-current slope of the ASE injection-locked WRC-FPLD almost remains as constant with varying ASE injection power, indicating that the reshaping on rising and falling edge by changing ASE injection power is negligible. If we fix the WRC-FPLD bias and the PRBS NRZ amplitude, the “off-state” level of upstream transmitted data will increase with larger ASE injection power, whereas the “on-state” level remains almost unchanged. Thus, the ER monotonically decreases with increasing ASE power when the WRC-FPLD enters into gain-saturation condition.
Fig. 9 SNR and ER of WRC-FPLD injection-locked by ASE with changing power levels in different systems.
In comparison with conventional WRC-FPLD based DWDM-PON, the use of additional 200 GHz AWG placed after the ASE source effectively suppress the crosstalk and interfered reflection between the upstream transmitted data and reflected ASE signal. Such a broadband reflection from between the based DWDM Mux and DeMux is greatly reduced from −14.7 to −21 dBm with a decreasing of 6.3 dB. We also find that the Schawlow-Towne’s linewidth (Δωst) becomes narrower in the new system even without SOA. Lower mode beating noise is obtained as the mode extinction becomes significant by reducing the ASE reflection level when comparing with conventional injection-locking scheme (with the ASE located in OLT) [14]. Due to the reduction of noise floor and mode beating noise, both the SNR and ER are improved from 5.9 to 6.1 dB and from 8.9 to 9.3 dB, respectively. It is clear seen that adding the SOA based noise blocker can further improve the SNR of upstream data by 0.3 dB. Moreover, the “off-state” level of upstream data will decrease by slightly increasing the threshold current of WRC-FPLD, such that the ER can essentially be ehanced from 9.3 to 9.6 dB. Apparently, the combining effect of the 200 GHz-AWG based spectral slicer and the SOA based noise blocker located after the ASE source essentially improves the ER from 8.9 to 9.6 dB and SNR from 5.9 to 6.3 dB. At ASE injection power of −3 dBm, the BER performances of the ASE injection-locked WRC-FPLD transmitter with three kinds of ASE sources located at different nodes are compared in Fig. 10 . At BER of 10−9, the newly proposed system shows a smallest receiving power sensitivity of −26 dBm, and there is a negative power penalty of 1.8 dB as compared to the conventional system in back-to-back case. After 25-km transmission in single-mode fiber, the receiving power of BER 10−9 degrades to −23.8 dBm, whereas the system withput SOA or the conventional system fails to achieve same BER level. The additional SOA filtering leads to a negative power penalty by 1 dB due to the improved SNR and ER under the reductions of crosstalk and mode beating noise.
Fig. 10 BER of filtered ASE with SOA (new system), filtered ASE (new system without SOA) and broadband ASE injection-locked WRC-FPLD based WDM-PON (inset: back-to-back eye diagram of (a) old system, (b) new system without SOA, (c) new system with SOA).
3. 3 Distinguished influence of SNR and ER to the BER performance of the WRC-FPLD transmitted up-stream data
To discriminate the individual contribution of ER and SNR to the BER, we adjust the ASE injection power and the WRC-FPLD biased current to detune ER and SNR of upstream transmitted data. By keeping the ER or SNR as constant, we investigate the effect of another parameter on the BER, as shown in Fig. 11 . Under the same ER of 10 dB, the receiving powers of BER at 10−9 are required to exceed −26 and −18 dBm at SNR of 6.3 and 5.3 dB, respectively. With constant SNR of 6.3 dB, the requested receiving powers of BER at 10−9 are enlarged from −26 to −22 dBm when ER is degraded from 10 to 8.6 dB. As a result, the sensitivity slopes for SNR and ER are determined as ΔPReceiver/ ΔSNR = 8.9 and ΔPReceiver / ΔER = 3.2, which clearly elucidate that the SNR plays a more important role than ER on BER performance.
Fig. 11 BER analysis of AWG-sliced and SOA-bleached ASE injection-locked WRC-FPLD transmitter with changing SNR (left) and changing ER (right).
Moreover, we also observe the biased current of WRC-FPLD is more pronounced than the injection power for ER and SNR of the upstream transmitted data. Apparently, the currents of SOA and WRC-FPLD are the most important parameters to achieve better transmission performance of the upstream data. In addition, it is worthy noting that the ASE channel linewidth filtered by AWG also affects the WRC-FPLD signal quality. In the experiment, the channel linewidth of 100 and 200 GHz AWG are 0.51 nm and 1.03 nm individually. In back-to back case, the BER analysis of upstream transmitted data through the AWG with different channel spacing is shown in Fig. 12 . When the 200GHz AWG is replaced by a 100 GHz one to filter ASE, the error floor of –log (BER) is degraded from 10.8 to 8.5. In the mean time, the SNR is degraded from 6.3 to 5.4 dB and ER is increased from 9.6 to 10 dB concurrently. In general, the EDFA or SOA based ASE light source exhibits two different noises, including the signal-spontaneous beat noise, and the spontaneous-spontaneous beat noise, which concurrently cause the degradation on the SNR of signals. In our case of the injection-locked WRC-FPLD with external ASE injection, the signal-spontaneous beat noise no longer sustains but the ASE-ASE beat noise injects into the WRC-FPLD and affects the noise performance of WRC-FPLD. The equation of SNRASE is defined as SNR = P2 ASE/(P2 sp-sp + P2 shot + P2 thermal) [15], where PASE is the ASE average power, Psp-sp is the spontaneous to spontaneous beat noise average power, Pshot is the shot noise average power, and Pthermal is the thermal noise average power. The shot noise and thermal noise are ignored as they are usually not in optical path. Due to the relationship of P2 sp-sp = 2P2 ASEBe/mΔλ, the SNR can be rewritten as SNR = mΔλ/2Be, where Be is the electrical bandwidth of the optical receiver, m is the number of polarization, and Δλ is the spectral linewidth of the ASE injection-locked WRC-FPLD transmitter. It is straight forward that the SNR as well as BER can be degraded by decreasing the spectral linewidth of the injected ASE source. Therefore, if we employ the AWGs with channel spacing of 100 GHz, it is clearly seen that the BER will be seriously degraded with a BER error floor as high as 4x10−7. If we further release the transmission channel linewidth in the DWDM-PON by using a 200 GHz AWG based Mux and Demux with spectral window of 1.0 nm (twice larger than that of a 100 GHz AWG based ones). The BER error floor of –log(BER) is greatly reduced from 6.5 to 8.4, while the SNR is enhanced from 4.2 to 5.5 dB and the ER is increased from 9.2 to 10 dB concurrently. In comparison, the broadening in ASE filter linewidth further improves the BER error floor of the AWG channelized and SOA filtered the ASE injection-locked WRC-FPLD upstream transmitted data by four orders of magnitude down to 10−11. That is, there is an extremely large power penalty up to 12 dB when reducing the ASE spectral linewidth from 1.0 to 0.5 nm before injection-locking the WRC-FPLD. This again corroborates that the SNR becomes the more pronounced effect than ER for promoting the BER performance. In addition, it is also concluded that the AWG channelized ASE injection-locked WRC-FPLD transmitter fails to promote the channeling capacity of the DWDM-PON system from 200 GHz to 100 GHz. Even the origined ASE spectral linewidth is broadened to 1.0 nm, the upstream transmission linewidth is still constrained by the AWG based Mux/Deux channel window. This eventually leads to the degradation of the receiving power sensitivity increasing from −27 to −15 dBm at BER if 10−9. From the ΔSNR, ΔER and BER of this experiment, we conclude the linewidth of AWG before WRC-FPLD has dominated effects on upstream signal quality, and the SNR plays a more important role than ER on the WRC-FPLD upstream data quality as well.
Fig. 12 BER of the WRC-FPLD transmitted data in the DWDM-PON systems with changing ASE-Demux-Mux linewidths of (a) 1.0-1.0-1.0 nm, (b) 0.5-1.0-1.0 nm, and (c) 0.5-0.5-0.5 nm.
4. Conclusion
By using a 200GHz AWG channelized ASE source in connection with a saturable semiconductor optical amplifier (SOA) based noise blocker as the injecting source at local ONU part, we demonstrate the spectrum-sliced ASE injection-locked WRC-FPLD transmitter directly modulated at 2.488 Gbit/s with greatly suppressed intensity noise performance in a 200GHz channelized DWDM-PON network. The experiments show that the low noise WDM-PON improves ER from 8.9 to 9.6 dB and SNR from 5.9 to 6.3 dB. In comparison with conventional (broad-band) ASE injection-locked WRC-FPLD transmitter at same power, there is an improvement on receiving power penalty (ΔPReceiver) by 2 dB at BER 10−9 in back-to-back case, and the receiving power of BER 10−9 can achieve −24 dBm even after 25km fiber transmission. Besides, with the additional AWG filtering, the crosstalk effect between the upstream transmitted data and the reflected ASE signal can be great reduced by 6.3 dB. The compromised effects of ER and SNR on BER performance are also elucidated via the modified SNR model for the WRC-FPLD under ASE injection induced gain-saturation condition. The ΔPReceiver/ΔSNR of 8.89 at same ER condition is more pronounced than the ΔPReceiver/ΔER of 3.17 obtained under same SNR condition, indicating that the SNR plays a more important role than the ER on enhancing the BER performance. In addtion, we research the impacts of the AWGs channel linewidth for the WDM-PON. From the results, we successfully find the currents of SOA and WRC-FPLD are the most important parameters for upstream signal and the linewidth of AWG before WRC-FPLD has larger effects than the AWGs after WRC-FPLD on upstream signal quality in this WDM-PON.
Acknowledgment
Financial support by the National Science Council of Taiwan Republic of China under grants NSC98-2221-E-002-023-MY3 and NSC 98-2623-E-002-002-ET are acknowledged.
References and links
1. H. D. Kim, S. Kang, and C. Lee, “A low-cost WDM source with an ASE injected Fabry–Pérot semiconductor laser,” IEEE Photon. Technol. Lett. 12(8), 1067–1069 (2000). [CrossRef]
2. S. L. Woodward, “P. P. lannone, K. C. Reichmann, and N. J. Frigo, “A spectrally sliced PON employing Fabry–Pérot lasers,” J. Lightwave Technol. 10, 1337–1339 (1998).
3. K.-Y. Park and C.-H. Lee, “Intensity Noise in a Wavelength-Locked Fabry–Perot Laser Diode to a Spectrum Sliced ASE,” IEEE J. Quantum Electron. 44(3), 209–215 (2008). [CrossRef]
4. A. McCoy, P. Horak, B. C. Thomsen, M. Isben, and D. J. Richardson, “Noise suppression of incoherent light using a gain-saturated SOA: Implications for spectrum-sliced WDM systems,” J. Lightwave Technol. 23(8), 2399–2409 (2005). [CrossRef]
5. S. Kim, J. Han, J. Lee, and C. S Park, “Intensity noise suppression in spectrum-sliced incoherent light communication systems using a gain-saturated semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 11(8), 1042–1044 (1999). [CrossRef]
6. M. Zhao, G. Morthier, and R. Baets, “Analysis and optimization of intensity noise reduction in spectrum-sliced WDM systems using a saturated semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 14(3), 390–392 (2002). [CrossRef]
7. X. Cheng, Y. J. Wen, Y. Dong, Z. Xu, X. Shao, Y. Wang, and C. Lu, “Optimization of Spectrum-Sliced ASE Source for Injection-Locking a Fabry–Pérot Laser Diode,” IEEE Photon. Technol. Lett. 18(18), 1961–1963 (2006). [CrossRef]
8. D. McCoy, B. C. Thomsen, M. Ibsen, and D. J. Richardson, “Filtering effects in a spectrum-sliced WDM system using SOA-based noise reduction,” IEEE Photon. Technol. Lett. 16(2), 680–682 (2004). [CrossRef]
9. G. P. Agrawal and N. A. Olsson, “Self-phase modulation and spectral broadening of optical pulses in semiconductor laser amplifiers,” IEEE J. Quantum Electron. 25(11), 2297–2306 (1989). [CrossRef]
10. G. P. Agrawal, Fiber-Optic Communication Systems, Third Ed., (Willy Inter-Science, 2002) Chaps. 4–6.
11. Y.-C. Chang, Y.-H. Lin, J. H. Chen, and G.-R. Lin, “All-optical NRZ-to-PRZ format transformer with an injection-locked Fabry-Perot laser diode at unlasing condition,” Opt. Express 12(19), 4449–4456 (2004). [CrossRef] [PubMed]
12. L. Li, “Static and dynamic properties of injection-locked semiconductor lasers,” IEEE J. Quantum Electron. 30(8), 1701–1708 (1994). [CrossRef]
13. K. Petermann, Laser Diode Modulation and Noise, (Publishers Dordrecht, The Nitherlands: Kluwer Academic, 1998), Chap. 2.
14. S. Gee, F. Quinlan, S. Ozharar, and P. Delfyett, “Two-mode beat phase noise of actively modelocked lasers,” Opt. Express 13(11), 3977–3982 (2005). [CrossRef] [PubMed]
15. J. S. Lee, Y. C. Chung, and D. J. DiGiovanni, “Spectrum-sliced fiber amplifier light source for mutichannel WDM application,” IEEE Photon. Technol. Lett. 5(12), 1458–1461 (1993). [CrossRef]
